8000 RFC: Getting rid of `numpy.f2py.compile()` · Issue #25122 · numpy/numpy · GitHub
[go: up one dir, main page]
Skip to content

RFC: Getting rid of numpy.f2py.compile() #25122

New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

Sign up for GitHub

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

Jump to bottom
Closed
HaoZeke opened this issue Nov 12, 2023 · 1 comment · Fixed by #25193
Closed

RFC: Getting rid of numpy.f2py.compile() #25122

HaoZeke opened this issue Nov 12, 2023 · 1 comment · Fixed by #25193
Labels
03 - Maintenance 15 - Discussion 54 - Needs decision component: numpy.f2py

Comments

@HaoZeke
Copy link
Member
HaoZeke commented Nov 12, 2023

Here is the implementation:

numpy/numpy/f2py/__init__.py

Lines 18 to 121 in 9340fca

def compile(source,
modulename='untitled',
extra_args='',
verbose=True,
source_fn=None,
extension='.f',
full_output=False
):
"""
Build extension module from a Fortran 77 source string with f2py.
Parameters
----------
source : str or bytes
Fortran source of module / subroutine to compile
.. versionchanged:: 1.16.0
Accept str as well as bytes
modulename : str, optional
The name of the compiled python module
extra_args : str or list, optional
Additional parameters passed to f2py
.. versionchanged:: 1.16.0
A list of args may also be provided.
verbose : bool, optional
Print f2py output to screen
source_fn : str, optional
Name of the file where the fortran source is written.
The default is to use a temporary file with the extension
provided by the ``extension`` parameter
extension : ``{'.f', '.f90'}``, optional
Filename extension if `source_fn` is not provided.
The extension tells which fortran standard is used.
The default is ``.f``, which implies F77 standard.
.. versionadded:: 1.11.0
full_output : bool, optional
If True, return a `subprocess.CompletedProcess` containing
the stdout and stderr of the compile process, instead of just
the status code.
.. versionadded:: 1.20.0
Returns
-------
result : int or `subprocess.CompletedProcess`
0 on success, or a `subprocess.CompletedProcess` if
``full_output=True``
Examples
--------
.. literalinclude:: ../../source/f2py/code/results/compile_session.dat
:language: python
"""
import tempfile
import shlex
if source_fn is None:
f, fname = tempfile.mkstemp(suffix=extension)
# f is a file descriptor so need to close it
# carefully -- not with .close() directly
os.close(f)
else:
fname = source_fn
if not isinstance(source, str):
source = str(source, 'utf-8')
try:
with open(fname, 'w') as f:
f.write(source)
args = ['-c', '-m', modulename, f.name]
if isinstance(extra_args, str):
is_posix = (os.name == 'posix')
extra_args = shlex.split(extra_args, posix=is_posix)
args.extend(extra_args)
c = [sys.executable,
'-c',
'import numpy.f2py as f2py2e;f2py2e.main()'] + args
try:
cp = subprocess.run(c, capture_output=True)
except OSError:
# preserve historic status code used by exec_command()
cp = subprocess.CompletedProcess(c, 127, stdout=b'', stderr=b'')
else:
if verbose:
print(cp.stdout.decode())
finally:
if source_fn is None:
os.remove(fname)
if full_output:
return cp
else:
return cp.returncode

It was lightly amended recently for meson use in #25111, however there is really very little justification for its (continued) existence. Some thoughts:

  • Things shouldn't be in __init__
  • It is essentially a wrapper around subprocess.run
  • Has barely been touched since it was brought in 18 years ago or so (from the git blame)
  • It leaks memory

The only slight functionality it provides is that it can work with a block of text (by generating a temporary file). Additionally, tests for compile are pretty broken on Windows, and given that it still generates a module file, and the fact that since the meson transition users cannot use most of the old -c commands, it is really easier to use externally or be explicit about using subprocess with environment variables set.

It has always been marked as experimental and subject to change, I would like to therefore remove it ASAP :)

Any user who needs such functionality would be better suited by writing their own thin wrapper over subprocess based functionality (or use something like sh).
@HaoZeke HaoZeke mentioned this issue Nov 12, 2023
Merged
@rgommers
Copy link
Member

Seems reasonable to me, but then again I'm not a user. Maybe write to the mailing list and do a bit of searching with a code search tool? If that doesn't yield objections, then I'd say go ahead.

@HaoZeke HaoZeke mentioned this issue Nov 20, 2023
Merged
@alphaparrot alphaparrot mentioned this issue Jun 19, 2024
Closed
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment
Assignees
No one assigned
Labels
03 - Maintenance 15 - Discussion 54 - Needs decision component: numpy.f2py
Projects
None yet
Milestone
No milestone
Development

Successfully merging a pull request may close this issue.

MAINT: Kill all instances of f2py.compile HaoZeke/numpy
2 participants
@rgommers @HaoZeke
0